Proefschrift
ter verkrijging van de graad van doctor aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben, voorzitter van het College van Promoties
in het openbaar te verdedigen op 6 juli in 2016 om 12:30 uur
door
Wouter VAN RIEL
ingenieur in integraal waterbeheer, Wageningen Universiteit geboren te Tilburg, Nederland
promotor prof. dr. ir. F.H.L.R. Clemens promotor prof. dr. ir. P.M. Herder copromotor dr. ir. J.G. Langeveld Samenstelling promotiecommissie:
Rector Magnificus voorzitter
Prof. dr. ir. F.H.L.R. Clemens Technische Universiteit Delft, promotor Prof. dr. ir. P.M. Herder Technische Universiteit Delft, promotor Dr. ir. J.G. Langeveld Technische Universiteit Delft, copromotor Onafhankelijke leden:
Prof. mr. dr. J.A. de Bruijn Technische Universiteit Delft Prof. dr. ir. A.R.M. Wolfert Technische Universiteit Delft
Dr. ir. F. Cherqui INSA de Lyon
Prof. dr. ir. W. Rauch University of Innsbruck
Prof. dr. ir. J.B. van Lier Technische Universiteit Delft, reservelid
Dit proefscrift is tot stand gekomen met ondersteuning van het Kennisprogramma Urban Drainage. De betrokken partijen zijn: ARCADIS, Deltares, Evides, Gemeente Almere, Gemeente Arnhem, Gemeente Breda, Gemeente ’s-Gravenhage, Gemeente-werken Rotterdam, Gemeente Utrecht, GMB Rioleringstechniek, KWR Watercy-cle Research Institute, Royal HaskoningDHV, Stichting RIONED, STOWA, Sweco, Tauw, vandervalk+degroot, Waterschap De Dommel, Waternet and Witteveen+Bos.
©2016 by W. van Riel ISBN: 978-94-6233-327-7
Printed by: Gildeprint, Enschede Cover design by A.A. van Riel
An electronic version of this document is available free of charge in the Delft University Repository at http://repository.tudelft.nl/.
I look back on roughly four years of joy. Joy in becoming a researcher who is al-lowed, even professionally obliged, to criticise everything around him. That is, work related of course. Joy in having numerous laughs with my colleagues. Joy in visiting all organisations participating in the research programme and discussing thoughts about current sewer management practices. Joy in communicating my results and ideas, obtained through uncommon approaches, to other researchers and managers in practice.
This joyful experience is in particular made possible by my supervisors: Jeroen Langeveld, Paulien Herder and Fran¸cois Clemens. You simply form an awesome team! I truly hope you will continue your cooperation, which proved to be fruitful. Each of you, but especially together, has the ability to energise me with your enthusiastic and critical attitude. Thank you for doing so.
My colleagues at Urban Drainage, thank you for the numerous hilarious nonsense discussions. Adithya, Antonio, Alex, Johan, Lisa, Marco, Mathieu, Matthieu, Nikola and Petra, thank you for your support, care, fun, discussions and game testing. Johan, I highly appreciate your sincere and good quality feedback whenever I asked for it or you found I needed it. I think you know almost as much about my topic as I do.
I would like to thank all parties participating in the ‘Kennisprogramma Urban Drainage’ for their contribution to my research. A special note of thanks goes to Irene Meyer for assisting in developing and programming Maintenance in Motion. Next to that, I would like to thank all players participating in Maintenance in Motion. The involved parties are: ARCADIS, Brabant Water, Deltares, Evides, Gemeente Almere, Gemeente Arnhem, Gemeente Breda, Gemeente Ede, Gemeente Utrecht, Gemeente-werken Rotterdam, Oasen, Royal HaskoningDHV, Samenwerking (Afval)waterketen Zeeland, Sweco, Vitens, Waterboard Brabantse Delta, Waterboard De Dommel and Waternet.
Wouter van Riel, 2016
Sewer systems are urban underground infrastructures for collecting and transporting wastewater and storm water to a treatment facility, which discharges it onto sur-face water. A properly functioning sewer system is important to society, because it provides two essential services: protection of public health and prevention of ur-ban flooding. As such, appropriate management of sewer systems, termed sewer asset management, is required in order to continue their service provision. An important as-pect within sewer asset management is decision-making for sewer replacement, which is the topic of this thesis.
Sewer asset management typically is a public responsibility. As such, one could expect the responsible organisations can justify their decision-making, and it is efficient and effective for lowest public costs (cost-effective). To this end, decision-making trans-parency is required. This seems difficult however, because sewer asset management is surrounded by technical and social complexity. This has two consequences. First, the difficulty to analyse, understand and predict structural condition and hydraulic sewer system performance. Second, difficulties in the many interactions between relevant stakeholders, their interests and negotiations between them. These effects decrease decision-making transparency, which is not preferred considering public accountabil-ity and cost-effectiveness. Next to that, the urban drainage sector typically assumes that better information about system performance, particularly structural condition, leads to better operational management.
Decision-making for a single actor is usually comparable with rational decision-making. This means: a sequential process of clearly defining a problem, obtaining informa-tion to weigh decision alternatives and making a decision. Objective informainforma-tion is important in this process. Multi-actor decision-making, also referred to as political decision-making, is different from the single actor process. The actors reason from their own interests, positions and values, and do not always have an equal objective. Hence, they may need to negotiate and make compromises in their decisions. Objec-tive information is less important here than in the rational process. The extent to which it is important is yet unknown.
This leads to the question, “Does higher quality information about structural con-dition lead to other or improved decision-making, when compared to the situation encountered in current practice?”
The objective of this thesis is to describe the actual decision-making processes and underlying information use in sewer asset management, in order to assess whether variations in information quality influence decision-making outcome. Interviews, a questionnaire and a serious game were applied for data collection, because the actual decision-making processes are not documented. Serious gaming was applied because it allows to study real actors, both individually and in groups, in a custom-built simulation environment.
The use of information and intuition in sewer replacement decisions was explored in a first round of interviews. Intuition is regarded as a success factor in decision-making in the perspective of ‘Naturalistic Decision Making’, or as a basis for bias in the perspective of ‘Heuristics and Biases’. The common ground is that intuitive decision-making stems from experience and cognitive pattern decision-making. Decision-decision-making for sewer replacement was found to involve the consideration of three aspects, each of which has multiple underlying information sources: the technical replacement need, the potential synergy from collaboration with other public works, and organisational preferences. A sewer asset manager appeared to combine these aspects in risk anal-yses, where particularly the probability of some effect, for example pipe collapse, showed to be estimated through intuitive judgments. It seems logical intuition is preferred to analytical reasoning, given the complex context the sewer asset manager operates in. Physical feedback of the sewer system to its manager, as a result of an applied replacement strategy, is hardly noticeable, because of the robustness of the sewer system itself the relatively long time it takes before this feedback occurs. Con-sequently, chances of learning from a chosen course of action are limited. As such, the preconditions for intuition to be skilled (sufficient regularity and opportunity to learn), are not met.
A second round of interviews extended this analysis by exploring the actual decision argumentation of executed sewer replacement projects in the Netherlands. Both the decision-making process and the content were analysed. Twenty-eight unique infor-mation sources were identified, which were found to be combined in a large variety of ways. Camera inspections, pipe age and planning of road works were mentioned most often. Approximately half of the projects was found to be initiated through a single actor decision-making process, which showed to resemble rational decision-making. The other half was initiated through a multi-actor decision-making process, showing characteristics of political decision-making. This means: multiple infrastructure man-agers cooperated, during which a mix of information and interests led to compromises about whether, where and how work could be executed simultaneously.
The results of the projects analysis did not reveal insights into how sewer asset man-agers mutually value the individual information sources and whether they share a frame of reasoning about this value. Therefore, a digital questionnaire was distributed to all Dutch municipalities. This questionnaire contained paired comparisons between the ten information sources mentioned most often in the projects analysis. Two as-pects were analysed: the perceived importance of information for hypothetical sewer replacement decisions and the extent to which the respondents are concordant with each other, i.e. a shared frame of reasoning. Camera inspection footages were found
to be valued most and pipe age was valued least. The respondents showed to be quite consistent per individual, implying their answers could be considered as reli-able. They showed to be concordant as a group, meaning that the respondents have a similar framework of reasoning for judging about the relative value of the infor-mation sources. The respondents had the opportunity to give feedback at the end of the questionnaire. This feedback revealed that most of the respondents found it difficult to make general comparisons without having a practical context. This indi-cates decision-making in practice may be steered by other mechanisms than purely combining information sources.
A serious game was built for further decision-making analysis, since decisions for sewer replacement are often initiated through multi-actor decision-making. This game, ‘Maintenance in Motion’, simulates single and multi-actor operational decision-making for infrastructure replacement. It aims to investigate the influence of infor-mation quality on replacement decisions, for single and multi-actor decision-making. To this end, four game types were designed. The game objective is addressed by analysing differences in game results between the game types, where the game types are:
single actor decision-making and perfect information about object state, single actor decision-making and imperfect information about object state, multi-actor decision-making and perfect information about object state, and multi-actor decision-making and imperfect information about object state. Players manage drinking water, gas, sewer or street infrastructures, where each infras-tructure was modelled as four equal and independent objects. They are challenged to manage their infrastructure as cost-effectively as possible, when playing individually. When playing as a group, they are challenged to balance their individual goal with their team goal (increasing overall infrastructure quality to minimize failure while minimizing overall public costs). The effect of the group effort, i.e. team utility, was measured by a different criterion than cost-effectiveness per individual, because the best strategy per actor depends on the choices of others. The game was designed as an experimental research instrument that allows for hypotheses testing concerning the relation between game outcome and player behaviour. Consequently, it was de-signed such that it had a relatively small solution space, measurable variables and a quantitative outcome analysis. As such, the final game model is a heavily simplified version of infrastructure management in reality.
The game was played with actual infrastructure managers in twenty-five sessions at eighteen different organisations. The game results proved to be valid and inferences could be made from them, based on analysis of the players’ applied management actions and answers to a questionnaire distributed after each gaming session.
Results from the gaming sessions show that when players were presented with per-fect instead of imperper-fect information about infrastructure condition, in a single actor decision-making environment, they managed their infrastructure more cost-effectively. The availability of perfect instead of imperfect information about infrastructure con-dition hardly changed game outcome in terms of team utility. Moreover, despite the availability of the team utility score, collaboration typically led to higher costs (ap-proximately 30 % on average) compared to the situation in which they would not collaborate. These results suggest that group choices are primarily based on negoti-ations that lead to compromises, instead of analytical reasoning as a group.
The presented results about actual decision-making for sewer replacement show a large gap between decision support models and decision-making in reality. Intuitive reasoning and dynamics of multi-actor decision-making are outside the scope of these models, but influence actual operational decision-making to a large extent. Increasing information quality about structural condition is only to a certain extent beneficial for increased cost-effective management. Such efforts would be meaningful particularly in single actor decision-making environments. As such, the current challenge for increased decision transparency and cost-effectiveness is unlikely to be solved by the current type of decision support tools for sewer asset management.
More research could be executed on development of multi-criteria decision support tools that can incorporate tacit next to explicit knowledge. Second, the underlying motivations of group decisions could be thoroughly examined, in order to understand why collaborative choices are made. If transparency is created in this respect, it allows for assessing whether and where decision-making may be improved. A new serious game could be developed that serves as a training tool, allowing players to experi-ence realistic feedback from their asset management strategies. Next to that, it may be possible to test and evaluate management strategies, alternative ways of budget allocation and organisational setups. Sewer asset managers in current practice could realise that neither every information source is relevant at all times, nor do these give perfect information. Sewer operators should be able to judge the impact of the uncer-tainties on the decisions they need to make on a day-to-day basis in order to motivate their choices properly, even in unpredictable decision-making processes.
Riolering is een belangrijke stedelijke infrastructuur die zorgt voor de inzameling van afval- en hemelwater en transport naar een zuiveringsinrichting, waarna het stedelijk afvalwater op oppervlaktewater wordt geloosd. Een goed werkende riolering is van groot maatschappelijk belang, omdat deze twee belangrijke diensten levert: bescherming van de volksgezondheid en ontwatering van het stedelijk gebied. Goed rioleringsbeheer is dan ook noodzakelijk om deze dienstverlening in stand te houden. Een belangrijke activiteit binnen het rioleringsbeheer is het besluiten tot rioolver-vanging, het onderwerp van dit proefschrift.
Over het algemeen zijn publieke organisaties verantwoordelijk voor rioleringsbeheer. Van deze organisaties mag men dan ook verwachten dat zij hun besluitvorming kunnen verantwoorden, en dat deze effici¨ent en effectief is voor zo laag mogelijke maatschap-pelijke kosten (doelmatig). Hiervoor is transparante besluitvorming noodzakelijk. Het is echter lastig om transparantie te verhogen omdat rioleringsbeheer omgeven is door technische en sociale complexiteit. Dit heeft de volgende effecten. Ten eerste is het moeilijk om systeemprestatie (hydraulisch en objecttoestand) te analyseren, te begrijpen en te voorspellen. Ten tweede is het moeilijk om de vele interacties tussen relevante actoren, hun belangen en onderlinge onderhandelingen te begrijpen. Deze effecten verlagen de transparantie van besluitvorming, hetgeen onwenselijk is in het kader van verantwoording van publiek bestuur en doelmatigheid. Daarnaast wordt in de rioleringssector veelal gedacht dat betere informatie over systeemprestatie, met name objecttoestand, leidt tot betere operationele besluitvorming.
Besluitvorming voor ´e´en actor lijkt veelal op rationele besluitvorming. Dit houdt in: een stapsgewijs proces van een eenduidig probleem defini¨eren, informatie vergaren om alternatieven te wegen en een besluit nemen. Objectieve informatie speelt hierin een belangrijke rol. Multi-actor besluitvorming, ook wel politieke besluitvorming ge-noemd, verloopt anders dan ´e´en-actor besluitvorming. De actoren redeneren vanuit verschillende belangen, posities en waarden, en hebben daarnaast niet altijd hetzelfde doel. Om deze redenen zijn de actoren genoodzaakt om te onderhandelen en com-promissen te sluiten wanneer zij samen besluiten willen nemen. De rol van objectieve informatie is hier een stuk kleiner dan in het rationele model. De grootte van deze rol is echter nog onbekend.
Dit leidt tot de vraag: “Zorgt kwalitatief betere informatie over objecttoestand voor andere of betere besluitvorming, vergeleken met de huidige situatie?”
Het doel van dit proefschrift is de werkelijke besluitvormingsprocessen en het on-derliggend gebruik van informatie te beschrijven voor rioolvervanging, om vervolgens na te gaan of variaties in informatiekwaliteit de uitkomst van besluitvorming be¨ınvloe-den. Er is gebruik gemaakt van interviews, een enquˆete en ‘serious gaming’, omdat informatie over de werkelijke besluitvormingsprocessen niet gedocumenteerd zijn. Se-rious gaming is toegepast omdat hiermee het gedrag van personen, zowel per individu als per groep, bestudeerd kan worden in een zelf gebouwde simulatieomgeving. In een eerste ronde interviews is een verkenning gemaakt van het gebruik van in-formatie en intu¨ıtie in rioolvervangingsbeslissingen. Intu¨ıtie wordt gezien als een succesfactor in het theoretisch kader van ‘Naturalistic Decision Making’ of als bron van fouten in het kader van ‘Heuristics and Biases’. Beide hebben gemeen dat intu¨ıtieve besluitvorming voortvloeit uit ervaring en cognitieve patroonherkenning. Besluitvorming voor rioolvervanging bleek gebaseerd te zijn op het in acht nemen van drie aspecten, die elk vervolgens zijn opgebouwd uit verschillende informatiebronnen. Deze drie aspecten zijn: de technische vervangingsbehoefte, potenti¨ele meerwaarde uit samenwerking met andere publieke werken, en organisatorische voorkeuren. Een rioleringsbeheerder combineert deze aspecten in risicoanalyses, waarin vooral de kans op een gevolg intu¨ıtief geschat wordt. Het bleek logisch dat intu¨ıtie wordt verkozen boven analytisch redeneren, vanwege de complexe context waarin de rioleringsbe-heerder werkt. Een fysieke terugkoppeling van de riolering naar de berioleringsbe-heerder, als gevolg van een toegepaste beheerstrategie, is nauwelijks merkbaar door de robuust-heid van de riolering zelf en de lange tijdsspanne waarover deze terugkoppeling zich over het algemeen manifesteert. De kans om iets te leren van het handelen wordt hiermee sterk verkleind. Om deze redenen wordt niet voldaan aan de gestelde voor-waarden (voldoende regelmaat en kansen om te leren) voor vakkundige intu¨ıtie. In een tweede ronde interviews is voorgaande analyse verder uitgebouwd via een verkenning van de beslissingsargumentatie van uitgevoerde rioolvervangingsprojecten in Nederland. Hierbij zijn zowel de inhoud als het proces van besluitvorming ge-analyseerd. Dit leidde tot achtentwintig unieke informatiebronnen die werden gecom-bineerd op veel verschillende manieren. Camera inspecties, buisleeftijd en planning van wegwerkzaamheden werden het vaakst genoemd. Ongeveer de helft van de on-derzochte projecten werd ge¨ınitieerd via een ´e´en-actor besluitvormingsproces. Deze processen leken op rationele besluitvorming. De andere helft van de projecten werd ge¨ınitieerd via een multi-actor besluitvormingsproces. Hierin bleken karakteristieken van politieke besluitvorming terug te komen. Dit hield in: samenwerking tussen twee of meer actoren, waarbij een mix van informatie en belangen leidde tot een onder-handelingsproces waarin compromissen werden gemaakt over of, waar en hoe werk gezamenlijk zou kunnen worden uitgevoerd.
De resultaten van voorgaande interviews toonden echter nog niet aan hoe riolerings-beheerders bronnen van informatie waarderen ten opzichte van elkaar en of zij een gedeeld denkkader hiervoor hebben. Om deze reden is een digitale enquˆete verspreid naar alle Nederlandse gemeenten met gepaarde vergelijkingen tussen de tien meest genoemde informatiebronnen uit voorgaande projectenanalyse. Twee aspecten zijn hierbij geanalyseerd: de ervaren waarde van informatie voor hypothetische
rioolver-vangingsbeslissingen en de aan- of afwezigheid van een gedeeld denkkader om deze waarde in te schatten. Camera inspecties werden als meest belangrijk ervaren en buisleeftijd het minst. De antwoorden van de respondenten zijn betrouwbaar, omdat zij behoorlijk consistent waren per individu. De respondenten bleken overeenstem-ming te hebben in de wijze waarop zij de informatiebronnen waardeerden. Dit houdt in dat een gedeeld denkkader aanwezig is. De respondenten hadden de mogelijkheid om feedback te geven na afloop van de enquˆete. Uit deze feedback bleek dat de respondenten het moeilijk vonden om generieke vergelijkingen te doen zonder infor-matie over de context te hebben. Dit impliceert dat besluitvorming in de praktijk waarschijnlijk gestuurd wordt door meer mechanismen dan puur het combineren van informatiebronnen.
Aangezien besluitvorming voor rioolvervanging deels afhangt van onderhoud aan an-dere infrastructuren, is voor veran-dere analyse een serious game gebouwd. Dit spel, ‘Maintenance in Motion’, simuleert ´e´en- en multi-actor operationele besluitvorming voor vervanging van infrastructuur. Het doel van dit spel is de invloed van infor-matiekwaliteit op vervangingsbeslissingen te kunnen onderzoeken, voor zowel ´e´en- als multi-actor omstandigheden. Om deze reden zijn vier speltypen gemaakt. Het spel-doel wordt bereikt door verschillen in spelresultaten tussen de speltypen te analyseren. Deze speltypen zijn:
´e´en-actor besluitvorming en perfecte informatie over objecttoestand, ´e´en-actor besluitvorming en imperfecte informatie over objecttoestand, multi-actor besluitvorming en perfecte informatie over objecttoestand, en multi-actor besluitvorming en imperfecte informatie over objecttoestand. Spelers beheren elk een drinkwater-, gas-, riolering- of weginfrastructuur, waarbij iedere infrastructuur gemodelleerd is als vier dezelfde en onafhankelijke objecten. De spelers worden uitgedaagd om een balans te vinden in hun individuele doel (doel-matigheid) en hun groepsdoel (verhogen kwaliteit infrastructuur en minimaliseren publieke kosten). Het nut van groepshandelen is beoordeeld met een ander criterium dan doelmatigheid, omdat de beste strategie per actor namelijk afhangt van de strate-gie van anderen. Het spel is ontworpen als een experimenteel onderzoeksinstrument om hypothesen te toetsten over de relatie tussen spelresultaten en gedrag van de spel-ers. Om die reden heeft het spel een relatief kleine oplossingsruimte, meetbare vari-abelen en een kwantitatieve resultatenanalyse. Het uiteindelijke spelmodel is daarom een behoorlijk versimpelde versie van infrastructuurbeheer in de praktijk.
Het spel is gespeeld met infrastructuurbeheerders uit de praktijk in vijfentwintig sessies bij achttien verschillende organisaties. Uit analyse van de toegepaste beheer-maatregelen van de spelers en de antwoorden op een enquˆete verspreid na afloop van iedere spelsessie bleek dat de spelresultaten valide zijn en conclusies kunnen worden getrokken op basis van de verkregen resultaten.
Resultaten van de spelsessies tonen dat spelers doelmatiger beheren, in ´e´en-actor situaties, wanneer zij perfecte in plaats van imperfecte informatie krijgen over de huidige objecttoestand. De beschikbaarheid van perfecte in plaats van imperfecte informatie over de huidige objecttoestand veranderde de uitkomst van het groepshan-delen nauwelijks. Ondanks de beschikbaarheid voor de spelers van informatie over het groepsresultaat leidde samenwerking in de meeste gevallen tot hogere kosten (gemid-deld circa 30 %), vergeleken met de situatie wanneer de spelers niet zouden samen-werken.
De gepresenteerde resultaten over besluitvorming voor rioolvervanging laten een groot gat zien tussen beslissingsondersteunende modellen en besluitvorming in de realiteit. Intu¨ıtie en de dynamiek van multi-actor besluitvorming vallen buiten het aandachtsveld van deze modellen maar be¨ınvloeden besluitvorming in sterke mate. Het verhogen van de nauwkeurigheid en betrouwbaarheid van informatie over sys-teemprestatie is slechts gedeeltelijk bevorderlijk voor de doelmatigheidsverhoging van de huidige beheerpraktijk. De meerwaarde van dergelijke inspanningen geldt vooral voor ´e´en-actor besluitvormingsprocessen. De huidige uitdaging voor verhoging van transparantie en doelmatigheid in rioleringsbeheer wordt daarmee waarschijnlijk niet behaald door de huidige typen beslissingsondersteunende modellen.
Onderzoek zou kunnen worden uitgevoerd naar een kosten-batenanalyse omtrent ver-hoging van informatiekwaliteit versus nut voor beheer. Ook zouden de onderliggende motivaties in multi-actor besluitvorming nader onderzocht kunnen worden om te be-grijpen waarom groepskeuzen gemaakt worden. Als meer transparantie hierin kan worden verkregen, kan worden nagegaan of, en waar, besluitvorming kan worden verbeterd. Een nieuwe serious game zou kunnen worden gebouwd voor trainings-doeleinden, waarin spelers realistische feedback krijgen van hun toegepaste beheer-strategie. Daarnaast kunnen verschillende beheerstrategie¨en, alternatieve wijzen van budgettoedeling en alternatieve organisatiestructuren worden getest en ge¨evalueerd. Rioleringsbeheerders in de praktijk zouden moeten beseffen dat geen enkele infor-matiebron altijd relevant of perfect is. Beheerders zouden het effect van onzekerheid moeten kunnen beoordelen om hun keuzes goed te motiveren, ook in onvoorspelbare besluitvormingsprocessen.
Acknowledgements vii
Summary ix
Samenvatting xiii
1 Introduction 1
1.1 Context and development of sewer asset management . . . 1
1.2 Current challenges . . . 4
1.3 Views on decision-making processes . . . 8
1.4 Objective, research questions and outline . . . 9
2 Individual decision-making: intuition versus information 13 2.1 Introduction . . . 13
2.2 Intuitive decision-making and relation with sewer asset management . 14 2.3 Research approach . . . 17
2.4 Results and discussion . . . 20
2.5 Conclusions and recommendations . . . 28
3 Individual and group decision-making: theory and practice 31 3.1 Introduction . . . 31
3.2 Rational versus political decision-making . . . 32
3.3 Research approach . . . 34
3.4 Results and discussion . . . 37
3.5 Conclusions and recommendations . . . 47
4 Individual decision-making: valuing information 49 4.1 Introduction . . . 49
4.2 Research approach . . . 50
4.3 Results and discussion . . . 54
4.4 Conclusions . . . 59
5 Individual and group decision-making: game setup 61 5.1 Introduction . . . 61
5.2 Serious games: what and why? . . . 62
5.3 Game design . . . 64
5.4 Game calibration and testing: methods . . . 77
5.5 Game calibration and testing: results and discussion . . . 81
5.6 Lessons learned and future research . . . 86
6 Individual and group decision-making: game results 89 6.1 Introduction . . . 89
6.2 Research approach . . . 91
6.3 Results and discussion . . . 96
6.4 Conclusions . . . 106
7 Discussion and concluding remarks 109 7.1 Discussion . . . 109 7.2 Conclusions . . . 111 7.3 Recommendations . . . 115 References 127 List of publications 129 Appendices
A List of information sources 131
B Costs as function of number of Monte Carlo simulations 133
C Game flowcharts 135
D Matlab script game engine 141
1.1
Context and development of
sewer asset management
Sewer systems are urban underground infrastructures for collecting and transport-ing sewage from houses and commercial buildtransport-ings, possibly mixed with excess storm water, to a treatment facility which discharges it onto surface water. Sewer sys-tems are critical infrastructures, because they provide essential services to society (Murray and Grubesic, 2007) protecting public health and preventing urban flood-ing. For example, the transition from cesspools to sewer systems triggered a decrease in death rate of approximately 30 (in 1850) to 8 (today) deaths per 1000 persons per year (Geels, 2006) (original in Mackenbach, 1992). Appropriate management of sewer systems, termed sewer asset management from here on, is required in order to continue their service provision.
The history of sewer systems goes back to the Indus Valley Civilisation, 3300-1300 BCE, during which covered drains were used to transported wastewater (Gray, 1940). It took until the 19th century before large scale sewer systems were constructed in European and American cities, with London being one of the first to undertake this effort, motivated by ‘the Great Stink’ in 1858 (Geels, 2006). Before that, cesspools and barrels were the primary faecal waste collection systems. Next to that, increased attention was paid to the relation between infectious diseases and exposure to drinking water contaminated with faecal bacteria (e.g. Snow, 1855). The effluent was typically discharged outside the urban area in natural streams. The transition to sewer systems proceeded until the 1930s. It continued after the Second World War, particularly during 1950-1970, as a result of the accelerated urbanisation. Wastewater treatment plants were built in order to reduce pollution load to surface waters.
Sewer asset management used to receive little attention. In the Netherlands, attention for sewer asset management started to increase in the mid-1980s. It is defined here as the set of activities required to maintain the performance of the system. Examples of such activities are cleaning, replacement, budget allocation and strategy formulation. System performance consists of two interrelated aspects:
Hydraulic performance, being the discharge and storage capacity of the system. Sufficient discharge capacity prevents urban flooding, whereas sufficient storage capacity limits emissions of wastewater to surface water (environmental perfor-mance). Acceptable environmental performance is a precondition for acceptable hydraulic performance.
Structural condition of the individual objects, including sewer pipes, gully pots and manholes.
Sewer system performance is typically maintained through repositioning, in areas with uneven soil settlement such as the west of the Netherlands, and replacing individual objects. The structural condition of individual pipes is rated in condition states by means of visual inspections (see CEN, 2011). Replacement is defined here as replace-ment of a pipe by another, irrespective of the newly installed pipe diameter.
Before the mid-1980s in the Netherlands, management practices had been about fix-ing incidental problems when noted, such as collapse of pipe segments, leakages, or malfunctioning for other reasons. Given the low frequency of problems, this approach was believed to be cost-effective. Yet, during the 1980s, the breakdown frequency increased considerably due to the use of inferior concrete and casting methods in the 1950s and 1960s. Next to that, many Dutch municipalities had insufficient available funds for replacement of sewers (adapted from Oomens, 1992). These developments raised public and political awareness of the need for a more systematic approach to sewer management (Thissen and Oomens, 1991). A major change was initiated in the Netherlands in 1993, because the ‘Environmental Management Act’ became ef-fective. This Act introduced the municipal duty of care for sewer systems and the obligation for municipalities to present a strategic municipal sewerage plan every five years. This plan describes policy objectives and costs for managing the sewer system. The costs for sewer asset management are fully covered by issuing taxes to house-holds and companies. The treatment of wastewater and the control of surface water quality are responsibilities of water boards, being independent governmental bodies. Both the duty of care and the legal obligation to present strategic sewerage plans is internationally unique.
From the 1980s to 2010, Dutch municipal management efforts focussed on two addi-tional aspects. First, areas outside city centres were connected through pressurised sewers. Approximately 22 % of the current sewer system length is of the pressurised type (RIONED Foundation, 2009). Second, after the 1990s, pollution load to surface water was preferred to be reduced further by increasing system storage capacity and changing system design from combined (one pipe for both wastewater and storm wa-ter) to separate (one pipe for wastewater and local infiltration and/or transport of storm water through pipes). Since the financial crisis in 2008, attention for system maintenance has grown further, focussing on explicit motivation of management
ef-forts. To this end, the concept of risk based decision-making was introduced, which includes additional aspects to base decisions on, for example health, liveability, sus-tainability, climate resilience and image (STOWA and RIONED Foundation, 2014). Yet, this conceptual approaches have not been adopted in current guidelines, which are described in the European Standard EN 752 (CEN, 2008) and, for the Nether-lands, in Urban Drainage Guidelines (in Dutch: Leidraad Riolering).
Almost 100 % of Dutch properties are currently connected to a sewer system, which is comparable to Germany, Switzerland and the UK. Other Western European countries have sewer connectivity rates of 76-95 % and Eastern Europe 50-75 %. Approximately 1.5 billion Euro per year is currently spent in the Netherlands for management of the 120,000 km of sewers (RIONED Foundation, 2013). This equals about 88 Euro/in-habitant/year. The largest portion of this budget, approximately 50-60 %, is deployed for sewer pipe replacement (obtained from multiple Dutch strategic municipal sewer-age plans). The annual sewer replacement rate in the Netherlands ranges between 1 and 2.5 %, depending on the assumed sewer lifetime. The annual inspection rate is approximately 10-15 %.
Sewer asset management was set up by adopting principles from systems theory and cybernetics for a structured approach. The principles of De Leeuw’s control paradigm (De Leeuw, 1974) were adopted (figure 1.1) in the work of Oomens (1992) to create a rational process model with all relevant activities for sewer asset management. The control paradigm is an open system model in which a controller, controlled system and environment interact, where the term open refers to interaction with an environ-ment.
The controller’s ability to successfully control its system depends on five preconditions for effective control.
1. The controller has an objective and an evaluation mechanism to check whether the goals are met.
2. The controller has a model of the controlled system to predict the effect of potential control actions.
3. The controller has information about the environment and the controlled sys-tem.
4. The controller has sufficient control actions to cope with the variability of the system.
5. The controller has sufficient information processing capacity to transform incom-ing information into effective control actions that are in line with the objectives. Adopting this rational process model can be seen as a logical choice at that time, because of the need for a structured and straightforward approach and the pragmatic application for municipalities (Oomens, 1992).
Controller Environment Controlled system Control Information Information Influence Output Influence
Figure 1.1: De Leeuw’s control paradigm (De Leeuw, 1974)
1.2
Current challenges
Sewer asset management typically is a public responsibility. As such, public account-ability applies, implying preconditions could be specified for the entities responsible for sewer asset management. These are: the obligation to explain and justify a course of action, and efficient and effective management at lowest public costs. Especially this latter precondition, defined as cost-effectiveness by Katz and Kahn (1978), has received increased importance the last decade. Decision-making transparency is re-quired in order to meet these preconditions. This is important because it could enhance the integrity of public governance, could improve performance, and provides managers and citizens with input for judging the fairness, effectiveness and efficiency of governance (Bovens, 2005) Decision-making for sewer asset management is, how-ever, not transparent enough to specify its level of cost-effectiveness or to assess whether, where and how it could be improved.
Decision-making transparency could be increased by addressing both substantive ra-tionality, the extent to which sound actions are chosen, and procedural rara-tionality, the effectiveness of the procedures used to choose actions (Simon, 1978, p. 9). Yet, decision-making for sewer systems, is embedded in a complex system, where com-plexity is defined as consisting of a high number of interacting physical and social elements (Bar-Yam, 1997; Sterman, 2000). In order to understand the behaviour of a complex system, it is important to understand not only the behaviour of the in-dividual parts, but also how they act together to form the behaviour of the whole (Bar-Yam, 1997; Bijker et al., 1987). This complexity can be separated in two types: system and process complexity.
System complexity refers to the many interactions among physical infrastructure com-ponents and their direct environment. Sewer systems are large underground networks consisting of pipes, manholes, gully pots, overflows and pumps. All these objects are used with a variety in shape and dimensions. The performance of a sewer system de-pends on joint and individual functioning of the objects in relation to the (movement of) surrounding soil, groundwater, surface water, other physical objects and human activities such as traffic movement. Even processes that occur within the pipe are to a large extent unknown, including sediment transport, fat accumulation and concrete degration over time due to biological composition and degradation of the wastewater (Ashley et al., 2004). As such, it may be considered logical that analysing and pre-dicting sewer system performance is difficult. The effect of this physical complexity is that many effects of these interactions, in terms of failure mechanisms, are largely unknown. That also counts for the relation between object and system failure. Phys-ical feedback of the sewer system about failure development is almost unobservable, because of the robustness of the sewer system and the relatively long time it takes before this feedback occurs. Given that sewer systems are underground, makes it even harder to observe or measure some aspects of system performance over time.
A challenge regarding system complexity is the difficulty to analyse and under-stand structural condition and hydraulic performance, despite the large number of studies devoted to this topic (e.g. Alegre, 2000; Ashley and Hopkinson, 2002; CEN, 2008; Carey and Lueke, 2013; Chughtai and Zayed, 2008; Egger et al., 2013; Fenner et al., 2000; Ferreira et al., 2011; Johansen et al., 2007; Kleidorfer et al., 2013; Le Gauffre et al., 2007; Marzouk and Omar, 2012; Matos et al., 2003; Sægrov, 2005; Scheidegger et al., 2011). First, the data required to accurately assess structural condition and predict failure rate, is not available. In turn, this diminishes the ef-fectiveness of prediction models for practical purposes. This data would support evaluation of management strategies and prediction of system performance to see whether potential control actions have any effect (preconditions 1, 2 and 3 of the control paradigm in figure 1.1). It has been believed, both in the Netherlands and elsewhere, that the necessary data to accurately evaluate and predict system per-formance and structural condition could be obtained. This requires prediction tools and data. Considering structural condition, available deterioration models are based on visual inspections and link system age to condition states (Baur and Herz, 2002; Egger et al., 2013; Scheidegger et al., 2011). A general disadvantage of visual inspec-tions is that it is difficult to convert observed defects on object scale to physical status and performance judgments on network to system scale. Second, time series are usu-ally unavailable, decreasing the predictive power of the observations. Third, visual inspection is limited to observations at the inner pipe. Fourth, human observations are prone to errors due to cognitive limitations in information processing, hampering repeatability (Dirksen et al., 2013). And fifth, information about the physical envi-ronment of sewer pipes is not included, but do influence the occurrence and severity of defects. Moreover, a link between system age and condition state is relatively mean-ingless if this relation excludes actual failure probabilities. Despite these drawbacks,
visual inspections are widely applied to assess pipe quality and serve as prime decision motivation (Van Riel et al., 2014b). In practice, this is essentially the only source of information sewer asset managers may use as a reference for decision support. Diffi-culties in assessing system performance result in an unclear relation between system performance and required sewer asset management. Consequently, the required sewer asset management effort to achieve a predefined service level is undefined.
Process complexity refers to the many interactions between relevant stakeholders and their interests. Decision-making in sewer asset management is a process of inter-action, in which the physical system boundaries are dynamic. Adjacent networked infrastructures are typically preferred to be rehabilitated simultaneously, motivated by reduction of costs, nuisance to citizens and traffic disruption. Figure 1.2 shows an example of a typical Dutch street cross section with several networked infrastruc-tures. The involved infrastructure managers are presented with a planning problem with a complex context (Parsons and Wooldridge, 2002; Rittel and Webber, 1973). The managers remain sovereign over their infrastructure, but negotiate and make compromises about whether, when, how and to what extent works are integrated. Although the initial motivations for joint rehabilitation seem relevant, the outcome of such multi-actor decision-making processes is usually unpredictable due to unpre-dictable human group behaviour. Decisions for system rehabilitation are influenced by interests of other actors in or outside the organisation the sewer asset manager operates. Examples are reputation issues towards citizens and politicians, political preferences for water management strategies, budget allocation or power and culture. Literature describing sewer asset management from an organisational decision-making perspective, including actor interaction aspects, is limited. Oomens (1992) is one of few sources that approaches sewer asset management from such an organisational perspective, describing a comprehensive overview of all activities of the sewer man-agement process. This description takes on a rather rational systems perspective for decision-making that is also applied to the interaction with other actors. As such, the influence of value trade-offs and intuitive decision-making is omitted in this work, which seem however, influential in sewer asset management (Johansen et al., 2007) or other sociotechnical complex systems (Gough and Ward, 1996; Westmacott, 2001).
Decision-making support, in the form of standards, guidelines or decision support models (including failure prediction and maintenance optimisation models), describe decision-making from a rational systems perspective (De Bruijn and Herder, 2009). These assistance tools propose to base decisions mostly on ‘objective’ information from the sewer system, for example pipe age, camera inspection images and hydraulic models. An example is shown in figure 1.3, depicting the process for sewer perfor-mance assessment from the European Standard EN752 ‘Drain and sewer systems outside buildings’ (CEN, 2008).
T re es Roa d E le ct ri ci ty T el ec om G as D ri nki ng w at er S torm s ew er F oul s ew er S tre et l ight ing P ri va te prope rt y bounda ry P ri va te prope rt y bounda ry G ul ly pot c onne ct ion H ous e c onne ct ion
Figure 1.2: Several networked infrastructures at a typical Dutch residential street
Results of investigation Assess hydraulic performance Assess environmental impact Assess structural condition Assess operational deficiencies
Compare with performance requirements
Identify hydraulic deficiencies Identify unacceptable environmental impact Identify structural deficiencies Identify operational deficiencies
Identify causes of performance deficiencies
Developing the plan
The sewer system management process in this standard has a rational character, depending on four information sources, which have to be compared with reference values to identify deficiencies and, subsequently, develop a plan. Interaction with other relevant actors, being part of the actor perspective (De Bruijn and Herder, 2009), is hardly addressed in the standard, because its focus is on objective information from the sewer system and its physical environment.
The consequence of this limited data availability and quality, complex context and negotiations between actors is that it hampers decision transparency. This leads to a situation where a fundamental question for sewer asset managers often remains unaddressed: why is a decision for implementing a measure actually taken? This notion undermines public accountability. In the urban drainage sector, it is typically assumed that extensive and good quality information about the system performance, especially structural condition, leads to better operational management. This leads to the question, does better information quality about structural condition (substantive rationality) lead to other or improved decision-making (process rationality), when compared to the situation encountered in present practice?
This question has been quantitatively addressed for individual decision-making (Chorus et al., 2007; Keller and Staelin, 1987), but not yet for multi-actor settings. On an individual level, increased quality of information about choice attributes increases the quality of decision outcome (Chorus et al., 2007; Keller and Staelin, 1987), but this effect decreases as the quantity of available information increases (Keller and Staelin, 1987). Group decision-making may outperform individual decision-making, because groups have advantages in terms of information processing and elimination of individual errors (Chalos and Pickard, 1985; Kocher and Sutter, 2005). On the other hand, group decision-making may suffer from ‘groupthink’, a psychological phenomenon triggering the individuals to seek harmony in a group decision-making process, although this harmony could lead to irrational outcomes (Janis, 1972). It is essentially unclear whether sewer asset management benefits from increased information quality about structural condition.
1.3
Views on decision-making processes
Decision-making theory generally distinguishes two main types of decision models: rational and political models. Decision-making in reality often has characteristics of both model types.
The rational model displays decision-making as a single actor oriented, stepwise and data driven approach. The involved decision-maker defines goals; defines alternative means for attaining them; evaluates the consequences of each alternative; and chooses the alternative most likely to attain the goal. In other words, the Homo economicus approach. In the course of time, the rational model of decision-making has undergone some modifications. Decision-makers consider only some alternatives, have limited information quantity and quality, and stop searching for a solution when they have found a satisfactory one for them, i.e. decision-makers simplify (Simon, 1955; Stone, 1988). Next to that, a rational process does not guarantee a rational outcome, be-cause substantive rationality may be limited (Simon, 1978), i.e. the needed data is unavailable or ambiguous.
The political view on decision-making emphasises that decisions are made by multiple actors. Actors are driven by different interests, e.g. following from their positions, roles, beliefs and values, which could change over time (March, 1994). Next to that, decision-makers often do not exactly know or have different perceptions about the problem and goal and the best way to reach it. Information processing is limited and consequences cannot be evaluated as unambiguously as suggested by the rational model (Etzioni, 1967; Lindblom, 1959). As a result, processes are less structured and staged than assumed by the rational model. The political view includes processes of cooperation, bargaining and making compromises which inevitably occur when more than one decision-maker is involved.
1.4
Objective, research questions and outline
Decision-making transparency is required considering public accountability and as-sessing cost-effectiveness. As such, this thesis aims at describing the actual processes and use of information in decision-making for sewer replacement, in order to assess whether variations in information quality influence decision-making. The focus is on decision-making for sewer replacement, because replacement works consume the largest portion of budget.
The aim is divided into the following fundamental research questions: 1. How does a sewer asset manager decide about sewer replacement?
2. How does a group of infrastructure managers decide upon joint public works? 3. How do variances in information quality influence decision outcome, both for
Case study research is chosen as approach because of the explorative character of this study. Interviews, questionnaires and serious gaming are used as data collection meth-ods, given the absence of documentation about actual decision-making processes for sewer replacement. A serious gaming approach was applied, because it is believed to be among the best methods for understanding complex systems, because it allows to incorporate real actors and interactions, physical rules, mental and computer models, and individual and collective goals (Bekebrede and Mayer, 2006, p. 278). The indi-vidual chapters elaborate on how each data collection method and analysis techniques were applied.
Figure 1.4 shows the relation between the three research questions and the applied methods to answer them.
Interviews:
• overview of information use and decision-making processes • executed replacement projects
Literature review: • decision-making theory • decision support models for
sewer asset management
Questionnaire: • assess importance of
information sources • assess shared framework of
reasoning
Serious gaming: • influence of information quality • influence of cooperation RQ 1 RQ 2 RQ 3 chapter 4 chapters 2 & 3 chapters 5 & 6
Chapter 2 describes sewer asset management from the perspective of the responsible management entity. It focuses on the use of objective information and intuition in decision-making for sewer replacement. The findings from chapter 2 are validated in chapter 3, which analyses the decision-making argumentation of executed sewer replacement projects. Emphasis is put on rational single actor versus political multi-actor decision-making. Chapter 4 elaborates on chapter 3 by assessing the relative importance of the consulted information sources. To this end, a digital questionnaire was sent to all Dutch municipalities. Then, chapter 5 presents the design considera-tions, setup and calibration of a serious game, ‘Maintenance in Motion’, to simulate single and multi-actor operational decision-making for infrastructure management. Chapter 6 presents the results obtained from the gaming sessions with infrastructure managers at Dutch municipalities, consultancy firms, drinking water companies and water boards.
intuition versus information
2.1
Introduction
Sewer systems are vital urban infrastructures, requiring appropriate asset manage-ment to safeguard serviceability and to balance service life versus costs for rehabili-tation. The basis of appropriate sewer asset management is sound decision-making, based on reliable data and information. Yet, decision-making for sewer asset manage-ment is inherently complicated, because it is embedded in a complex socio-technical system. Sewer asset managers face this complexity through various constraints, de-scribed in section 2.2.
These constraints force sewer asset managers to make intuitive decisions to fit the situation at hand, diminishing transparency of the decision-making process. Intuition is similar to expert judgment, both using tacit knowledge as a basis formed by expe-riences. Intuition is regarded as a success factor in decision-making in the concept of ‘Naturalistic Decision Making (NDM)’ (Klein, 2008), or as basis for bias in the con-cept of ‘Heuristics and Biases (HB)’ (Tversky and Kahneman, 1974). The common ground is that intuitive thinking and decision-making stems from experience and cog-nitive pattern making (Gobet and Chassy, 2009; Simon, 1983; Zsambok and Klein, 1997). The judgments and decisions called intuitive come to mind on their own, without explicit awareness and without an explicit evaluation of the validity. A fire fighter feels that a house is dangerous and a chess player sees a promising move (Kahneman and Klein, 2009, p. 519). When judgments about sewer systems are based on tacit knowledge, it is difficult to recall their underlying trade-offs and argumenta-tions and, considering investments, whether budgets are deployed properly.
This chapter is based on: Van Riel, W., Langeveld, J.G., Herder, P.M. & Clemens, F.H.L.R., 2014. Intuition and information in decision-making for sewer asset management. Urban Water Journal, 11 (6):506-518.
Still, as intuition is to a large part based on experiences and cognitive pattern making, the question is whether sewer asset managers base their decisions on ‘relevant’ expe-riences and patterns, and therefore employ skilled or flawed intuitive thinking in such socio-technical systems. It is unclear which and how information sources are used in deciding upon sewer system rehabilitation, and to what extent replacement decisions are influenced by intuition. This observation is not essentially different from any other decision-making process, but in order to improve current sewer asset management, this study assesses the availability and use of information in decision-making for sewer system replacement.
2.2
Intuitive decision-making and relation with
sewer asset management
Intuition is a topic studied intensively for decades, emerging from the fields of philos-ophy and psychology (e.g. Jung, 1921). Intuitive decision-making is deeply embedded in the evolutionary history of Homo Sapiens, being probably the most dominant mode of risk assessment and survival (Slovic et al., 2004). Due to its influence in decisions, it cannot be omitted from decision-making analysis. Two aspects about intuition are considered important for this study: conditions under which intuition is preferred over analytical reasoning, and circumstances influencing the success of intuitive decision-making.
Simon (1947) related intuitive thinking to organisational behaviour and concluded that the perfectly rational Homo economicus has to be replaced by a man of lim-ited knowledge and information processing capacity, Homo stultitia, i.e. rationality is bounded. Simon (1992, p. 155) describes intuition as decision-making behaviour that is speedy and for which the expert is unable to describe in detail the reasoning or other process that produced the answer, being “nothing more and nothing less than recognition”. To the view of Simon (1983), intuition is assembled through learning and experience and stored in long-term memory. Simon stressed the advantages of intuitive decision-making, which were later incorporated in the approach of ‘Natural-istic Decision Making (NDM)’ that emerged in 1989 (Klein, 2008). NDM postulates that under conditions of time pressure, ambiguity and changing conditions experts can make good decisions without having to consciously perform extensive, multi-attribute analyses. This is explained by the ‘recognition-primed decision (RPD) model’ (Klein, 1989; Zsambok and Klein, 1997) describing that people are able to make successful intuitive decisions by employing their experience to recognise problems as similar to problems previously experienced. An opposite view was initiated by the ‘Heuristics and Biases (HB)’ approach by Tversky and Kahneman (1974), who showed systematic cognitive biases accrue from reliance on judgmental heuristics, caused by a number of fallacies and miscomputations inherent in human information processing. These intuitive judgments arise from simplifying heuristics, not from specific experience. Consequently, such intuitive judgments are less likely to be accurate and are prone to systematic biases (Kahneman and Klein, 2009, p. 519). The common thought of the
NDM and HB approaches is that, regarding dual process models, intuitive judgments are produced by ‘system 1 operations’ in the brain, which are automatic, involuntary, and almost effortless. In contrast, the deliberate activities of ‘system 2 operations’ are controlled, voluntary and effortful, demanding cognitive effort. These two opposite views towards intuition pose the question of how skilled intuition can be distinguished from heuristic-based intuition (Kahneman and Klein, 2009).
When is intuition likely to be used in decision-making? Agor (1986) surveyed 200 ‘highly intuitive’ managers on issues related to the use of intuition. He found that the conditions under which intuition ‘functioned best’ included:
uncertainty,
absence of precedent,
requirement to use limited or ambiguous data and information, existence of equally plausible alternatives, and
time pressure.
Similar characteristics were later described by (Orasanu and Connolly, 1993, p. 7), listing eight factors that give rise to the use of intuitive judgments.
III-structured problems. The decision problem does not present itself in a neat and complete form, resulting in no single or correct answer.
Uncertain dynamic environments. Decision-making takes place in a world of incomplete and imperfect information and changing environments.
Shifting, ill-defined, or competing goals. The decision-maker is expected to be driven by multiple purposes, not all of them clear, some of which will be opposed to others.
Action/feedback loops. The traditional decision models are concerned with an event, a point in time at which the single decisive action is chosen. In contrast, it is much more common to find an entire series of events, a string of actions over time that are intended to deal with the problem, or to find out more about it, or both. Action/feedback loops may also generate problems. Actions taken and results observed may be only loosely coupled to one another, making it hard to attribute effect to cause.
Time stress. Decision-makers in these settings will often experience high lev-els of personal stress, with the potential for exhaustion and loss of vigilance. Second, their thinking will shift, characteristically in the direction of using less complicated reasoning strategies (Payne et al., 1988).
Multiple players. Many problems of interest involve not a single decision-maker, but several, perhaps many, individuals who are actively involved in one role or another. It can be hard to make sure all team members share the same understanding of goals and situational status so that relevant information is brought forward when needed in the decision process.
Organisational goals and norms. The organisational setting is relevant to the decision-making process in two ways. First, the values and goals that are being applied will not be simply the personal preferences of the individuals involved. Second, the organisation may respond to the decision-maker’s various difficulties by establishing more general goals, rules, standard operating procedures, service doctrine, or similar guidelines.
When is intuition considered to be skilled? Skilled intuition was defined by Simon (1992, p. 155) as “nothing more and nothing less than recognition”. According to Kahneman and Klein (2009, p. 520), the recognition model implies two conditions must be satisfied for an intuitive judgment to be genuinely skilled. First, skilled intu-itions will only develop in an environment of sufficient regularity, which provides valid cues to the situation. For instance, poker or chess games. In these situations, a rela-tion can be observed between a decision and the effect of the decision (performance). How does this relate to sewer asset management? For several decisions, a relation between decision and performance can easily be observed. For example, changing the hydraulic properties of a sewer system by decreasing the diameter of several pipes or closing a combined sewer overflow. This will inevitably change the hydraulic perfor-mance, being noticed at the first rain event. For system replacement however, such a relation between decision and performance is absent, because the time between a re-placement decision and its effect exceeds the professional life of sewer asset managers. Second, in case sewer pipes are replaced before their technical end of life, the replaced pipes’ condition is not checked and to what extent replacement is required.
A second condition for intuition to be skilled is that people must have an adequate opportunity to learn the relevant cues. For example, a chess player or musician requires years of deliberate practice to get skilled (Ericsson, 2006). Related to sewer asset management, learning is meant to take place through evaluation of the applied replacement strategy. Yet, the evaluation step is absent in European legislation (see CEN, 2008). In practice, the effect of a replacement strategy is not evaluated, resulting in a limited learning opportunity.
Similar to skilled intuitions, incorrect intuitions also arise from memory, and are caused for example by inadequately checking intuitive choices, or by attribute substi-tution (i.e. a difficult question is replaced by an easier one, while they do not have a high correlation) (Kahneman and Klein, 2009, p. 522). As such, incorrect intu-itions are likely to develop in ‘wicked’ situations (see Rittel and Webber, 1973), due to limited regularity (Hogarth, 2001). Moreover, professionals could develop overcon-fidence about the accuracy of their judgments, leading to an illusion of skill (Arkes, 2001).
A way to augment professional judgment is by the use of algorithms or other forms of decision support tools. Yet, these must remain under adequate human supervision, to provide monitoring of their performance. This is however difficult, because people tend to get more passive and less critical when such tools are in charge without giving feedback. This is defined as automation bias (Skitka et al., 1999).
Overall, intuitive decision-making is fast, because the brain draws conclusions by recognition from a few observations. Intuition, tacit knowledge and non-conscience pattern-making assist humans as their ‘mental butlers’, because these require little cognitive effort (Bargh and Chartrand, 1999, p. 476). It can provide successful deci-sions in professional contexts in complex conditions, under circumstances of regularity and opportunities to learn.
2.3
Research approach
The aim of this chapter is to increase understanding of the decision-making process regarding sewer system replacement. Case study research seems appropriate for this purpose, given the explorative and practical character of this study.
2.3.1
Case studies and data collection
The decision-making process and current use of information was assessed at seven municipalities in the Netherlands. Eighteen interviews were conducted at seven Dutch municipalities, ranging in population size from approximately 50,000 to over 750,000 inhabitants. These municipalities constitute approximately 15 % of total population (Statistics Netherlands, 2012) and approximately 9 % of the total sewer length in the Netherlands (RIONED Foundation, 2009). Table 2.1 shows several characteristics of the municipalities included for the study in this chapter.
Table 2.1: Characteristics of included municipalities
Municipality Nr. of inhabitants at 01-01-20111 Population density2 Sewer length∗ Available budget for 2012∗ Available budget per inhabitant Available budget per km sewer pipe (-) (inh./km 2
land) (km) (M Euro) (Euro/inh.) (k Euro/km)
Almere 190,655 1,469 1,100 8.7 45.6 7.9 Amsterdam 779.808 4,700 3,811 64.9 83.2 17.0 Barneveld 52,490 298 624 9.1 173.4 14.6 Breda 174,599 1,379 1,050 13.5 77.3 12.9 Ede 108,285 340 986 9.6 88.7 9.7 Rotterdam 610,386 2,987 2,906 51.2 83.9 17.6 The Hague 495,083 6,046 1,439 33.3 67.3 23.1
∗Data is extracted from the strategic municipal sewerage plan per municipality 1
(Statistics Netherlands, 2012)
2
Determine renewal need of system
Controlled system to controller
Determine which system parts are considered for renewal
Determine which system parts have priority in renewal
Localize other public works
Localize other public works
Environment to controller
Set up and prepare renewal project
Set up and prepare renewal project
Execute renewal project
Execute renewal project Determine renewal projects Determine renewal projects
Determine which system parts are considered for renewal at locations from step 2 Determine renewal need of system at locations from step 1
1 2 3 4 5 6 7
Figure 2.1: Flow charts used during interviews
The interviewees were selected as follows: first the heads of the ‘sewer system de-partments’ were selected based on their willingness to participate in an interview and asked to participate. Second, the department heads appointed one or two additional interviewees (dependent on the municipality’s staff size), requested by the authors, whom they thought to be relevant to be interviewed.
An in-depth (face to face) and semi-structured interview setup was chosen to collect data. This approach was chosen, because of the explorative character of the study and the complexity of the subject. It allows flexibility during the interview regarding structure, content and questions, which are framed by a network of topics that will be addressed. Semi-structured interviews suit the explorative character of this study. The in-depth interview was chosen, because of the complexity of the subject and to allow maximum diversity in the responses.
The interviews were guided by the flowcharts in figure 2.1. These were used to assist in discussing the decision-making process and information use. The flowcharts had been prepared based on the results of four exploratory interviews with experts in the urban drainage sector. The starting point of each flowchart differs, related to information flow in the control paradigm. As depicted in section 1.1, two information flows can be distinguished, which relate to external of internal impulse for system replacement: from controlled system to controller and from the environment to controller.
Through the flowcharts, the following topics were addressed in chronological order: justification and completeness of flowcharts, information sources per step, budget allocation and identification of organisational levels. The interviewees described these topics in a general sense, based on their knowledge and experience.
Seventeen of eighteen interviews were digitally recorded, approved by the interviewees. Afterwards, the interview recordings were fully transcribed. The transcript of the not recorded interview was sent to the interviewee for review.
Dutch municipalities are legally obliged to present a strategic municipal sewerage plan every five years, which are often publicly available. Each of these plans contains a specific section about decision argumentation for sewer replacement. Data was extracted from these sections.
2.3.2
Data analysis
Three types of data were analysed: interview data about the decision process, in-terview data about the use of information in the decision process, and data about the use of information in the decision process from the strategic municipal sewerage plans.
The first data source is analysed by open coding, both in-vivo and descriptive. The open coding approach was sufficient, because the objective was to describe the decision process by using preliminary flow charts.
The second data source, interview data about the use of information and intuition in the decision process, is analysed by content analysis. This is defined as “a research technique for making replicable and valid inferences from texts (or other meaningful matter) to the contexts of their use” (Krippendorff, 2004, p. 18). The steps for content analysis involve coding, making categories of the codes and abstraction, with the purpose of describing a phenomenon (Elo and Kyng¨as, 2008). This type of analysis is usually appropriate when existing theory or research on a phenomenon is limited. A specific type of content analysis, summative content analysis, was applied in this study. This type starts with identifying and quantifying words and contents with the purpose of exploring usage, and adds the underlying meaning of these words and contents (Hsieh and Shannon, 2005, p. 1283). Typically, the keywords used for coding are identified before and during data analysis, derived from interests of researchers or literature review (Hsieh and Shannon, 2005, p. 1286). Summative content analysis fits more to the objective of this study than regular content analysis, because of study’s focus on assessing usage of information and intuition in the decision process of sewer system replacement. The following steps were taken.
1. Open coding. The interview transcriptions were screened and in-vivo and scriptive coding was applied to the words and phrases to indicate each de-scribed information source and intuitive judgment used in the decision process. The identified keywords and phrases for the analysis of information usage were derived from literature and interviews with experts in the urban drainage sec-tor. The use of intuition was analysed by coding expressions reflecting intuitive thinking processes, including ‘feeling’, ‘interpretation’, ‘common sense’ and ‘in-tuition’. All codes were counted afterwards.
2. Axial coding. The interrelationships between the codes were identified in order to make code groups and categories.
